For CRO teams

Treat UGC like an experiment, not a vibe.

Run UGC placements against a holdout, read significance before you ship, and attribute the lift, all without a Core Web Vitals regression.

Get the lift teardown Open the CRO toolset

In good company

Sample workspace

CRO teams

idukki · live

Production · eu-west-2

37KB
runtime
1.2s
LCP p75
99.95%
uptime
SOC 2
in prog

Live
▶@maya · #ootd
£89
Shoppable video
PDP · 6 SKUs
<Widget />
Live
★★★★★4.9 · 2,310
- Best fit · runs cool
- Sized true to size
- Shipped in 2 days
✓ VerifiedJSON-LD ✓
Reviews wall
Homepage · live
<Widget />
Live
UGC gallery
PLP · 24 posts
<Widget />
Live
Story 2 of 5
Story rail
Mobile · sticky
<Widget />

+Drop in another widget

No code

+22%
Avg incremental PDP CR lift
Illustrative cohort midpoint; rerun on your own holdout
95%
Significance bar
Welch z-test + sequential MAB before auto-promote
37 KB
Widget runtime
gzipped, lazy-loaded media, no CWV regression
4 minutes
Time to first test

Why CRO teams pick Idukki

Conversion teams don’t want another widget. You want a clean test: a control, a variant, a significance bar you trust, and a revenue number you can defend. The reason UGC is worth a test slot at all is in the data, video reviews convert 4.1× better than text-only on the same SKU (PowerReviews, 2023) and 79% of shoppers say UGC influences what they buy (Nosto). Idukki runs those placements in shadow mode against a holdout, measures incremental lift per layout, and pipes the events into GA4 and Meta CAPI, all on a 37 KB widget that lazy-loads so the PDP speed score holds.

Test placements against a control
Run a UGC layout against a true holdout in shadow mode. Read incremental lift with a Welch z-test and sequential MAB, and only promote the winner once it clears significance. No eyeballing a dashboard and calling it a win.
Lift the PDP without a CWV hit
The widget is a 37 KB ESM bundle that lazy-loads media, so LCP, CLS and INP hold while you add the social-proof block. You get the conversion lever without paying for it in Core Web Vitals.
Attribute revenue per variant
Per-layout, per-variant revenue as first-class GA4 and Meta CAPI events, plus Klaviyo blocks. The lift number that lands in your experiment readout is the same one Finance can reconcile.
Migrate without downtime
Run Idukki alongside your incumbent first, then cut over in a single day with reviews, UGC and rights state preserved. No flicker, no gap in the funnel, no customer-facing downtime.

Proof, same role, real customers

The modules you’ll use most

FAQ

How do you measure incremental lift rather than correlation?
You run the UGC placement against a holdout in shadow mode and read the delta with a Welch z-test (and a sequential MAB when you want to allocate traffic as evidence accrues). Promotion only happens once the result clears your significance bar, so the number you report is incremental, not a coincidence on a good week.
Will adding the widget regress Core Web Vitals?
No. The runtime is 37 KB gzipped and lazy-loads media, so LCP, CLS and INP hold. Run the /tools/widget-speed checklist against your current setup if you want the before/after on your own pages.
Can we run it alongside our current tool during the test?
Yes. Idukki runs in shadow mode next to any incumbent, so you can validate lift before committing. Migration is a single-day cut-over with reviews, UGC and rights state preserved, and no customer-facing downtime.

Talk to us